|
Pareto analysis evolutionary and learning systems
Yaochu JIN, Robin GRUNA, Bernhard SENDHOFF
Front Comput Sci Chin. 2009, 3 (1): 4-17.
https://doi.org/10.1007/s11704-009-0004-8
This paper attempts to argue that most adaptive systems, such as evolutionary or learning systems, have inherently multiple objectives to deal with. Very often, there is no single solution that can optimize all the objectives. In this case, the concept of Pareto optimality is key to analyzing these systems. To support this argument, we first present an example that considers the robustness and evolvability trade-off in a redundant genetic representation for simulated evolution. It is well known that robustness is critical for biological evolution, since without a sufficient degree of mutational robustness, it is impossible for evolution to create new functionalities. On the other hand, the genetic representation should also provide the chance to find new phenotypes, i.e., the ability to innovate. This example shows quantitatively that a trade-off between robustness and innovation does exist in the studied redundant representation. Interesting results will also be given to show that new insights into learning problems can be gained when the concept of Pareto optimality is applied to machine learning. In the first example, a Pareto-based multi-objective approach is employed to alleviate catastrophic forgetting in neural network learning. We show that learning new information and memorizing learned knowledge are two conflicting objectives, and a major part of both information can be memorized when the multi-objective learning approach is adopted. In the second example, we demonstrate that a Pareto-based approach can address neural network regularizationmore elegantly. By analyzing the Pareto-optimal solutions, it is possible to identifying interesting solutions on the Pareto front.
References |
Related Articles |
Metrics
|
|
Evolutionary multi-objective optimization:some current research trends and topics that remain to be explored
Carlos A. COELLO COELLO
Front Comput Sci Chin. 2009, 3 (1): 18-30.
https://doi.org/10.1007/s11704-009-0005-7
This paper provides a short review of some of the main topics in which the current research in evolutionary multi-objective optimization is being focused. The topics discussed include new algorithms, efficiency, relaxed forms of dominance, scalability, and alternative metaheuristics. This discussion motivates some further topics which, from the author’s perspective, constitute good potential areas for future research, namely, constraint-handling techniques, incorporation of user’s preferences and parameter control. This information is expected to be useful for those interested in pursuing research in this area.
References |
Related Articles |
Metrics
|
|
Integrated biological systems modeling: challenges and opportunities
Jialiang WU, Eberhard VOIT
Front Comput Sci Chin. 2009, 3 (1): 92-100.
https://doi.org/10.1007/s11704-007-0011-9
Most biological systems are by nature hybrids consist of interacting discrete and continuous components, which may even operate on different time scales. Therefore, it is desirable to establish modeling frameworks that are capable of combining deterministic and stochastic, discrete and continuous, as well as multi-timescale features. In the context of molecular systems biology, an example for the need of such a combination is the investigation of integrated biological pathways that contain gene regulatory, metabolic and signaling components, which may operate on different time scales and involve on-off switches as well as stochastic effects. The implementation of integrated hybrid systems is not trivial because most software is limited to one or the other of the dichotomies above. In this study, we first review the motivation for hybrid modeling. Secondly, by using the example of a toggle switch model, we illustrate a recently developed modeling framework that is based on the combination of biochemical systems theory (BST) and hybrid functional Petri nets (HFPN). Finally, we discuss remaining challenges and future opportunities.
References |
Related Articles |
Metrics
|
|
Optimisation of algorithm control parameters in cultural differential evolution applied to molecular crystallography
Maryjane TREMAYNE, Samantha Y. CHONG, Duncan BELL
Front Comput Sci Chin. 2009, 3 (1): 101-108.
https://doi.org/10.1007/s11704-009-0009-3
Evolutionary search and optimisation algorithms have been used successfully in many areas of materials science and chemistry. In recent years, these techniques have been applied to, and revolutionised the study of crystal structures from powder diffraction data. In this paper we present the application of a hybrid global optimisation technique, cultural differential evolution (CDE), to crystal structure determination from powder diffraction data. The combination of the principles of social evolution and biological evolution, through the pruning of the parameter search space shows significant improvement in the efficiency of the calculations over traditional dictates of biological evolution alone. Results are presented in which a range of algorithm control parameters, i.e., population size, mutation and recombination rates, extent of culture-based pruning are used to assess the performance of this hybrid technique. The effects of these control parameters on the speed and efficiency of the optimization calculations are discussed, and the potential advantages of the CDE approach demonstrated through an average 40% improvement in terms of speed of convergence of the calculations presented, and a maximum gain of 68% with larger population size.
References |
Related Articles |
Metrics
|
|
Incorporating prior knowledge into learning by dividing training data
Baoliang LU, Xiaolin WANG, Masao UTIYAMA
Front Comput Sci Chin. 2009, 3 (1): 109-122.
https://doi.org/10.1007/s11704-009-0013-7
In most large-scale real-world pattern classification problems, there is always some explicit information besides given training data, namely prior knowledge, with which the training data are organized. In this paper, we proposed a framework for incorporating this kind of prior knowledge into the training of min-max modular (M3) classifier to improve learning performance. In order to evaluate the proposed method, we perform experiments on a large-scale Japanese patent classification problem and consider two kinds of prior knowledge included in patent documents: patent’s publishing date and the hierarchical structure of patent classification system. In the experiments, traditional support vector machine (SVM) and M3-SVM without prior knowledge are adopted as baseline classifiers. Experimental results demonstrate that the proposed method is superior to the baseline classifiers in terms of training cost and generalization accuracy. Moreover,M3-SVM with prior knowledge is found to be much more robust than traditional support vector machine to noisy dated patent samples, which is crucial for incremental learning.
References |
Related Articles |
Metrics
|
|
Evolutionary algorithm based on schemata theory
Takashi MARUYAMA, Eisuke KITA
Front Comput Sci Chin. 2009, 3 (1): 123-129.
https://doi.org/10.1007/s11704-009-0001-y
The stochastic schemata exploiter (SSE), which is one of the evolutionary algorithms based on schemata theory, was presented by Aizawa. The convergence speed of SSE is much faster than simple genetic algorithm. It sacrifices somewhat the global search performance. This paper describes an improved algorithm of SSE, which is named as cross-generational elitist selection SSE (cSSE). In cSSE, the use of the cross-generational elitist selection enhances the diversity of the individuals in the population and therefore, the global search performance is improved. In the numerical examples, cSSE is compared with genetic algorithm with minimum generation gap (MGG), Bayesian optimization algorithm (BOA), and SSE. The results show that cSSE has fast convergence and good global search performance.
References |
Related Articles |
Metrics
|
|
symbolic model checking APSL
Wanwei LIU, Ji WANG, Huowang CHEN, Xiaodong MA, Zhaofei WANG
Front Comput Sci Chin. 2009, 3 (1): 130-141.
https://doi.org/10.1007/s11704-009-0003-9
Property specification language (PSL) is a specification language which has been accepted as an industrial standard. In PSL, SEREs are used as additional formula constructs. In this paper, we present a variant of PSL, namely APSL, which replaces SEREs with finite automata. APSL and PSL are of the exactly same expressiveness. Then, we extend the LTL symbolic model checking algorithm to that of APSL, and then present a tableau based APSL verification technique, which can be easily implemented via the BDD based symbolic approach. Moreover, we implement an extension of NuSMV, and this adapted version supports symbolic model checking of APSL. Experimental results show that this variant of PSL can be efficiently verified. Henceforth, symbolic model checking PSL can be carried out by a transformation from PSL to APSL and symbolic model checking APSL.
References |
Related Articles |
Metrics
|
12 articles
|